Affective polarization — the dislike, distrust, and avoidance of those who hold different political views — has been one of the defining features of American public life over the past two decades. It shows up in who Americans choose to befriend, what causes families to rupture, and what college students feel free to say in class.

It makes the work of higher education harder: harder for students to learn from peers who see the world differently, harder for faculty to teach openly on politically charged topics, and harder for institutions to sustain the kind of open inquiry their missions are built on. Affective polarization is also a key part of a broader pattern that emerges when trust and engagement break down: lower intellectual humility during disagreement, greater misperceptions of what people with opposing views actually believe, and a deeper reluctance to speak openly about one’s views.

A growing number of researchers and practitioners have converged on a response: promoting constructive dialogue. Constructive dialogue is a form of conversation in which people with different perspectives seek to understand one another, without abandoning their own beliefs, in order to live, learn, and work together. The aim is not to change anyone's politics. It is to make engagement across political differences possible — and, where appropriate, productive.

Dialogue programs in their current form — brief and scalable — are more recent, and the evidence supporting them was, until recently, not as strong as their claims.

Even now, when one considers the psychological complexity of a constructive engagement between two people with different views, a skeptic may reasonably ask: Can you actually measure constructive dialogue?

Here we show how the Constructive Dialogue Institute, and the field in general, are tackling this question, and what has been discovered so far.

Why Colleges Need Better Dialogue Measurement

Building a campus culture that fosters constructive dialogue is not a task any one program can achieve on its own. As we note elsewhere (see CDI's Five Pillars Model), it depends on institutional leaders, norms and policies, curricula, and the interactions of everyday life: what students actually experience in classrooms, residence halls, and student organizations. Programs that focus on mindsets and skillsets are one piece of a larger project. This is why CDI has developed facilitation training programs for faculty and student leaders that reach the classroom and residential dorms, as well as a leadership institute where university leaders can formalize their commitment to fostering constructive dialogue through policies, third spaces, and more. But crucially, none of the pieces of this project can be meaningfully improved without a valid way to measure what is actually changing.

This is the case for measurement in constructive dialogue. The best way to know whether an institution is moving the needle on dialogue is by having a needle in the first place. A decade ago, an organization could win grants and institutional buy-in by describing a compelling theory of change and pointing to testimonials. The bar has changed, and both the science of constructive dialogue and our institutions are better for it.

Measurement allows us to compare different psychological interventions that seek to promote dialogue apples-to-apples. A vivid example is the Stanford-led Strengthening Democracy Challenge, which tested 25 interventions head-to-head on a standard set of outcomes, including partisan animosity (affective polarization), with more than 32,000 American partisans. The study gave institutions and funders a clear basis for comparing interventions on their impact, while producing some of the most credible estimates yet of what these programs can accomplish. CDI applies this same standard to its own work, tracking a defined set of outcomes across our programs — which we turn to now.

Key Outcomes for Constructive Dialogue

If the goal is to establish cultures of constructive dialogue, one key question is: what student-level outcomes drive culture change? The field has converged on a set of answers in recent years, which can be organized roughly into three categories:

(i) how students think about disagreement (i.e., affective polarization and intellectual humility)

(ii) what they can actually do in difficult conversations (i.e.,negative conflict behaviors and finding common ground)

(iii) how they experience the campus around them (i.e., sense of belonging)

Together, these reflect essential signals to capture for measuring constructive dialogue. We define each here.

• Affective Polarization — The extent to which people dislike and distrust those with opposing political views. e.g., “How do you feel about individuals who identify as liberal/progressive?”

• Intellectual Humility — Awareness of one’s cognitive limitations and openness to the possibility of being wrong. e.g., “I accept that my beliefs and attitudes may be wrong.”

• Finding Common Ground — The tendency to identify shared experiences, values, identities, and goals with those who have different views. e.g., “If I were to have a disagreement with someone, I believe I could effectively seek and build on areas of agreement between myself and the other person.”

• Negative Conflict Behaviors — The tendency to respond to conflict with negative/hostile behaviors. e.g., “If I were to have a disagreement with someone, I'm willing to speak in a disrespectful manner, depending on how the other person acts.”

• Sense of Belonging — How much a person feels they belong in a group. e.g., “I feel like I belong at my college/university"

Each connects to constructive dialogue in a specific way. Affective polarization is a central focus of research and a key emotional barrier to dialogue: students high on this measure are more likely to distrust and avoid those on the other side of a political issue, making dialogue harder to initiate and sustain. Intellectual humility, by contrast, is viewed as a cognitive facilitator of dialogue: dialogue requires openness to learning from the other side, and intellectual humility helps make that openness possible.

Finding common ground is a core dialogue capacity: the tendency to seek out shared values, concerns, or experiences with people one disagrees with — even when those commonalities are not initially apparent. Negative conflict behaviors, by contrast, capture what students tend to do when disagreement actually arises — these are a key behavioral outcome that dialogue programs are trying to change.

Belonging is a climate factor with decades of empirical grounding in educational psychology, where it has been shown to predict student retention and persistence. In the context of dialogue, students who feel they cannot hold or share their views without social cost are less likely to speak, and a campus in which most students stay silent is not a culture where dialogue can meaningfully happen.

Together, these five outcomes give the field a workable definition of what to look for. They can be measured with validated scales, compared across studies and institutions, and collected at the scale that campus-wide measurement requires.

To probe affective polarization, for example, researchers commonly use the "feeling thermometer": students rate how warmly they feel toward conservatives and progressives on a 0-to-100 scale (from very cold to very warm). A student's polarization score is the gap between how warmly the student rates their own side and the other. This “difference score” approach has proven useful for capturing the relational nature of polarization: it tells us about the size of the gap between feelings toward “us” and “them”, whether it’s a difference of 5 (outgroup) and 55 (ingroup) or 25 (outgroup) and 75 (ingroup).

Of course, the outcomes above are not the only ones that matter for polarization and dialogue, and not every intervention should be expected to move all of them. Programs target different parts of the problem, and meaningful evaluation depends on tracking the right outcomes for the right intervention. Ideally, these outcomes are measured over time as well, to track whether initial gains hold or grow over time.

Lastly, these quantitative measures are often complemented by qualitative measures: open-ended questions and interviews that let students explain, in their own words, what worked and why. In our recent report, Engaging Students Across Difference, we draw on interviews with students, faculty, and dialogue practitioners to identify four principles for designing effective dialogue interventions: that they be practical, reflective, scaffolded, and low-stakes. Findings like these offer guidance for the field in refining interventions and the outcomes we measure — and they would not have surfaced from quantitative measures alone.

What the Research Says About Constructive Dialogue

"How do we know whether what we're doing on campus is actually working?" This is a question countless administrators invested in dialogue and civic engagement have echoed in recent years.

It is a crucial one, and answering it is becoming more possible each year as established measures from social, moral, and political psychology are applied with increasing rigor. One recent effort shows what rigorous measurement can reveal when applied at scale.

In the Strengthening Democracy Challenge (noted above), a group of Stanford researchers put out an open call to identify effective interventions for reducing partisan animosity and strengthening democratic commitments. 252 interventions were submitted, and 25 were chosen by an expert panel to test. Of those selected, 23 of the 25 interventions significantly reduced partisan animosity. An abridged version of CDI's learning program, Perspectives, was one of them. Notably, no single program emerged as the answer — instead, a variety of interventions produced positive change. This shines light on another benefit of measurement: the opportunity to iterate over time towards what demonstrably works. For instance, these empirical findings prompted us to incorporate aspects of other successful interventions into our own Perspectives program, further strengthening its potential impact.

The efficacy of Perspectives was further supported by a peer-reviewed study published in 2023, which examined CDI's full Perspectives program across three studies, including two randomized controlled trials. In a sample of 775 college students randomly assigned to Perspectives or a waitlist control, students who completed the program showed meaningful reductions in affective polarization, gains in intellectual humility, and reductions in negative conflict behaviors.

The takeaway thus far: constructive dialogue can be measured, and we can learn which interventions are effective. This, in turn, can inform the design of stronger interventions in the future.

The Future of Measuring Constructive Dialogue

Most of what the field measures today is not dialogue itself, but the individuals who participate in it: their attitudes, reported behaviors, and sense of belonging on campus. That work has produced important insights. But the question of measuring constructive dialogue goes beyond any individual. It points to the conversation itself — the back-and-forth between two people on a contested issue, and whether that exchange has the qualities that we consider constructive.

This is where the field is going next. Large language models are increasingly being deployed to code natural language at scale for many features that used to require slow human rating. Moreover, a recent line of computational work has identified linguistic markers of conversational receptiveness — the use of language that communicates one’s willingness to thoughtfully engage with opposing views. These new tools show great promise to help us understand what features of a conversation form the necessary ingredients to make engagement across differences productive.

A second frontier moves from individual conversations to campus climate. Campuses already generate measurable signals related to dialogue: course evaluation items on viewpoint engagement, campus climate surveys, participation records for deliberative events, and repeated measures of classroom expression or self-censorship. Integrating these signals with survey-based outcomes could help institutions assess dialogue climate without relying exclusively on student surveys.

Both research directions are still in their early stages, but the emerging answer to our opening question is yes — we can increasingly measure constructive dialogue, and at increasing depth. Today’s measures show what is changing in the people who participate in our interventions. The next generation of methods will show, with growing clarity, what is changing in the conversations themselves — and in the campuses around them.